On macOS and Windows there are several ways to point an AI at something on your screen. On Linux the usual answer is still: take a screenshot, open a browser tab, upload it. LuciQA is a desktop app that does the select-and-ask step natively on Linux as well.
What it does
LuciQA sits in the system tray. Choose Ask from text (OCR) or Ask from image / diagram, drag over part of the screen, and the answer appears in a small panel beside your selection.
- OCR runs on your machine. Text is recognized locally in nine languages, and only the recognized text is sent.
- Image mode sends the selected region as it is, for charts, formulas and diagrams.
- Your own API key. Gemini, Groq, OpenRouter, Cloudflare and Mistral have free tiers. There is no LuciQA account, and requests go directly from your computer to the provider.
- Keys are stored in the desktop keyring (the Secret Service API, for example GNOME Keyring), not in a plain text file.
What you need
- A 64-bit (x64) Linux desktop.
- X11 or XWayland. LuciQA uses the X11 backend to place the selection overlay. On a Wayland session it runs through XWayland; a Wayland-only session without XWayland is not supported.
- A system tray. On GNOME that means an extension such as AppIndicator support.
- A keyring service that is unlocked when you log in.
Installing
There are two packages on the download page: an AppImage and a tar.gz. Neither is signed, so compare the SHA-256 shown on the page with the file you downloaded.
On most distributions the AppImage only needs to be made executable:
chmod +x LuciQA-*-linux-x64.AppImage
./LuciQA-*-linux-x64.AppImage
Ubuntu 24.04 and later restrict unprivileged user namespaces, which every Chromium-based app’s sandbox relies on. There an AppImage also needs FUSE 2 (libfuse2t64) and an AppArmor permission for its exact path. Install the AppImage to a root-owned path and load the profile that ships inside it:
sudo apt-get install libfuse2t64
sudo install -d -o root -g root -m 755 /opt/luciqa-appimage
sudo install -o root -g root -m 755 LuciQA-*-linux-x64.AppImage /opt/luciqa-appimage/LuciQA.AppImage
/opt/luciqa-appimage/LuciQA.AppImage --appimage-extract luciqa-appimage.apparmor
sudo install -o root -g root -m 644 squashfs-root/luciqa-appimage.apparmor /etc/apparmor.d/luciqa-appimage
sudo apparmor_parser -r /etc/apparmor.d/luciqa-appimage
/opt/luciqa-appimage/LuciQA.AppImage
The profile allows user namespaces for that one executable only; the system-wide restriction stays on. Do not start LuciQA with --no-sandbox to get around this.
What is still rough
The current state, plainly:
- No distribution package yet. There is no .deb, Flatpak or AUR entry, so installation is manual.
- Updates on a root-owned install are manual. The app tells you a new version exists and opens the download; you replace the file yourself.
- It is not open source. The app is free to use with your own keys, and Pro is a one-time purchase, but the code is closed.
- Wayland screen capture goes through the desktop portal, which asks for permission and behaves differently between desktops.
If one of these stops you, tell us which distribution and desktop you use. That is the most useful report a Linux user can send right now.